Spatial Transcriptomics Analysis of Vulnerable Cell Populations in Middle Temporal Gyrus

Abstract

AbstractBackgroundAlzheimer’s disease (AD) is characterized by cognitive decline due to histopathological changes of neurodegeneration in a stereotypical neuroanatomical progression. Applying single‐nucleus ‘omics methods has extended our understanding of how AD impacts the brain and allowed the Seattle AD Brain Cell Atlas Center (SEA‐AD) to provide a detailed cellular‐molecular view of middle temporal gyrus (MTG) in AD. Previously, we identified transcriptomically‐defined cell types that are specifically affected by AD in MTG, however, cell types identified in ‘omics data lack spatial context. Identifying the location of these cell types in situ is necessary to pinpoint the location of affected cells and connect cellular‐molecular views of disease progression to histopathology. To accomplish this, spatial transcriptomics methods measure individual mRNA molecule locations with subcellular resolution and provide transcriptomic profiles of individual cells in situ. MethodUsing a panel of genes designed to identify transcriptomic cell types, we performed spatial transcriptomics experiments using the Vizgen MERSCOPE platform across MTG tissue from 25 donors spanning AD pathology, yielding transcriptional profiles of over 1M cells across the cortical depth and white matter of 68 sections. These cells were assigned to transcriptomically‐defined cell types by integrating the spatial transcriptomics data with our existing single nucleus dataset. From the resulting maps of cell type locations, we measured cortical layer thickness, cell proportions, densities and cell type‐dependent pairwise distance distributions for each donor.ResultWe have successfully applied the MERSCOPE platform to generate cellular‐resolution spatial transcriptomic data in MTG across a wide range of AD disease progression. We applied novel deep generative models to assign cell type identity. These measurements revealed cell type‐specific spatial distributions and changes in cell type proportions due to disease, validating observations from single‐nucleus studies. Enhanced loss of Layer 2/3 excitatory neuron types and somatostatin‐expressing inhibitory neurons is calculating cell density within layers. Additional analyses uniquely available to spatial transcriptomics studies will be useful for quantitative description of cytoarchitectural changes in AD.ConclusionSpatial transcriptomics experiments across disease progression in MTG confirm cell type abundance changes measured in single‐nucleus RNASeq. The cellular spatial distributions provide detailed knowledge of the location of vulnerable cells within cortex.